Synthesis of extended transaction models using ACTA
ACM Transactions on Database Systems (TODS)
An overview of workflow management: from process modeling to workflow automation infrastructure
Distributed and Parallel Databases - Special issue on software support for work flow management
Managing heterogeneous multi-system tasks to support enterprise-wide operations
Distributed and Parallel Databases - Special issue on software support for work flow management
Workflow handbook 1997
Post-Capitalist Society
A Transactional Model for Long-Running Activities
VLDB '91 Proceedings of the 17th International Conference on Very Large Data Bases
Specifying and Enforcing Intertask Dependencies
VLDB '93 Proceedings of the 19th International Conference on Very Large Data Bases
Hi-index | 0.00 |
In a large organization, at any moment there may exist a large number of process instances. If a failure occurs a considerable number of them may be affected. Thus enhancing reliability is an important issue in workflow applications. Traditionally, reliability is enhanced by replication. In the context of business processes with a client-server architecture, this can be done by replicating process instances at multiple servers. However, because of the large number of process instances and the potentially large amount of information pertaining to each instance, maintaining consistency among replicas is more complex and hence causes in a business environment a higher overhead than do traditional data-intensive applications. Thus the use of replication for business processes should be limited. In this paper, we propose a scheme to enhance reliability of business process without replication. The scheme incurs lower overhead than any replication scheme, yet can improve the reliability over the normal non-replication case. Our scheme can be used as a complement to replication. For example, it can be adopted in a case where the degree of criticality of a process instance does not warrant its replication, yet improving its reliability is desirable. Another merit of the scheme is that recovery can start even before the failed server is repaired.